基于神经网络的制冷模糊控制系统仿真与优化

Simulation and optimization for fuzzy system of refrigeration based on neural networks

针对制冷系统被控对象模型建立了一个基本模糊控制器,并给出一组较优量化因子,在此基础上建立和训练了一个模糊神经网络控制器,并在MATLAB仿真环境下对制冷模糊控制进行仿真和优化。根据得到的一系列仿真结果对所设计的制冷模糊控制系统性能进行了分析,探讨了神经网络训练样本数据对FNNC系统控制性能的影响,结果表明采用合理的样本数据可以优化模糊神经网络制冷控制系统。

In this paper, in accordance with the mathematical model of Refrigeration system, a common fuzzy controller is created firstly. Then a group of quantitative factor with better control behavior is offered, on the base of which a Fuzzy Neural Networks Controller（FNNC） is created and trained. The FNNC is then used to control the Refrigeration system in the MATLAB environment and with a group of simulation results as well as according analytic data the author＇ s comments on the advantages of ANNC is given.